What Kagglers Are Using for Text Classification

Re:What Kagglers Are Using for Text Classification by biomodel on Thursday December 27, 2018 @11:49AM (#18770601)

Not sure why anyone would use 2D CNNs for processing text when there is no spatial correlation in the embedding features. Recent work such as https://arxiv.org/abs/1803.01271 show that for most tasks, 1D CNNs outperform recurrent architectures while being faster to train

[ Parent | on HN ]

Re:What Kagglers Are Using for Text Classification by madavidj on Thursday December 27, 2018 @04:45PM (#18772890)

Probably because the author followed this blog: http://www.wildml.com/2015/12/implementing-a-cnn-for-text-cl...

That blog used a 2d cnn because tensorflow didn't have a 1d version at the time of writing, so he just created a dummy 2nd dimension of length 1 and called it a day.

[ Parent | on HN ]

Re:What Kagglers Are Using for Text Classification by soraki_soladead on Thursday December 27, 2018 @01:50PM (#18771521)

This is just a bug in their code. The paper they cite uses 1D convolutions. Though, I suppose having an unused dimension only really hurts efficiency.

[ Parent | on HN ]

Re:What Kagglers Are Using for Text Classification by gnulinux on Thursday December 27, 2018 @05:00PM (#18772997)

> Though, I suppose having an unused dimension only really hurts efficiency.

That might not be true as it might increase bias and thus might need a more careful hyperparameter tuning to avoid overfitting.

[ Parent | on HN ]